Index: kernel/arch/ia64/src/ivt.S =================================================================== --- kernel/arch/ia64/src/ivt.S (revision 2277e03b97a6950736870a381ad825874af27449) +++ kernel/arch/ia64/src/ivt.S (revision 824208bff6a427e9769eab137e526ce9915f925f) @@ -31,15 +31,10 @@ #include #include +#include #include -#define FRS_TO_SAVE 30 -#define STACK_ITEMS (21 + FRS_TO_SAVE * 2) -#define STACK_FRAME_SIZE ALIGN_UP((STACK_ITEMS * STACK_ITEM_SIZE) + STACK_SCRATCH_AREA_SIZE, STACK_ALIGNMENT) - -#if (STACK_ITEMS % 2 == 0) -# define STACK_FRAME_BIAS 8 -#else -# define STACK_FRAME_BIAS 16 -#endif +#define STACK_FRAME_SIZE ALIGN_UP(ISTATE_SIZE + STACK_SCRATCH_AREA_SIZE, STACK_ALIGNMENT) + +#define FLOAT_ITEM_SIZE (STACK_ITEM_SIZE * 2) /** Partitioning of bank 0 registers. */ @@ -120,9 +115,9 @@ mov r27 = ~PSR_DFH_MASK ;; and r26 = r25, r26 - and r24 = r24, r27;; - or r24 = r24, r26;; - mov psr.l = r24;; + and r24 = r24, r27 ;; + or r24 = r24, r26 ;; + mov psr.l = r24 ;; srlz.i - srlz.d;; + srlz.d ;; mov r24 = cr.iip @@ -161,6 +156,6 @@ (p4) mov r12 = R_KSTACK ;; - add r31 = -STACK_FRAME_BIAS, r12 ;; - add r12 = -STACK_FRAME_SIZE, r12 + add r12 = -STACK_FRAME_SIZE, r12 ;; + add r31 = STACK_SCRATCH_AREA_SIZE + ISTATE_OFFSET_IN6, r12 /* 4. save registers in bank 0 into memory stack */ @@ -179,22 +174,22 @@ */ -(p6) st8 [r31] = r38, -8 ;; /* save in6 */ -(p6) st8 [r31] = r37, -8 ;; /* save in5 */ -(p6) st8 [r31] = r36, -8 ;; /* save in4 */ -(p6) st8 [r31] = r35, -8 ;; /* save in3 */ -(p6) st8 [r31] = r34, -8 ;; /* save in2 */ -(p6) st8 [r31] = r33, -8 ;; /* save in1 */ -(p6) st8 [r31] = r32, -8 ;; /* save in0 */ -(p5) add r31 = -56, r31 ;; +(p6) st8 [r31] = r38, -STACK_ITEM_SIZE ;; /* save in6 */ +(p6) st8 [r31] = r37, -STACK_ITEM_SIZE ;; /* save in5 */ +(p6) st8 [r31] = r36, -STACK_ITEM_SIZE ;; /* save in4 */ +(p6) st8 [r31] = r35, -STACK_ITEM_SIZE ;; /* save in3 */ +(p6) st8 [r31] = r34, -STACK_ITEM_SIZE ;; /* save in2 */ +(p6) st8 [r31] = r33, -STACK_ITEM_SIZE ;; /* save in1 */ +(p6) st8 [r31] = r32, -STACK_ITEM_SIZE ;; /* save in0 */ +(p5) add r31 = -(7 * STACK_ITEM_SIZE), r31 ;; - st8 [r31] = r30, -8 ;; /* save old stack pointer */ - - st8 [r31] = r29, -8 ;; /* save predicate registers */ - - st8 [r31] = r24, -8 ;; /* save cr.iip */ - st8 [r31] = r25, -8 ;; /* save cr.ipsr */ - st8 [r31] = r26, -8 ;; /* save cr.iipa */ - st8 [r31] = r27, -8 ;; /* save cr.isr */ - st8 [r31] = r28, -8 ;; /* save cr.ifa */ + st8 [r31] = r30, -STACK_ITEM_SIZE ;; /* save old stack pointer */ + + st8 [r31] = r29, -STACK_ITEM_SIZE ;; /* save predicate registers */ + + st8 [r31] = r24, -STACK_ITEM_SIZE ;; /* save cr.iip */ + st8 [r31] = r25, -STACK_ITEM_SIZE ;; /* save cr.ipsr */ + st8 [r31] = r26, -STACK_ITEM_SIZE ;; /* save cr.iipa */ + st8 [r31] = r27, -STACK_ITEM_SIZE ;; /* save cr.isr */ + st8 [r31] = r28, -STACK_ITEM_SIZE ;; /* save cr.ifa */ /* 5. RSE switch from interrupted context */ @@ -204,7 +199,7 @@ mov r26 = cr.ifs - st8 [r31] = r24, -8 ;; /* save ar.rsc */ - st8 [r31] = r25, -8 ;; /* save ar.pfs */ - st8 [r31] = r26, -8 /* save ar.ifs */ + st8 [r31] = r24, -STACK_ITEM_SIZE ;; /* save ar.rsc */ + st8 [r31] = r25, -STACK_ITEM_SIZE ;; /* save ar.pfs */ + st8 [r31] = r26, -STACK_ITEM_SIZE /* save ar.ifs */ and r24 = ~(RSC_PL_MASK), r24 ;; @@ -231,8 +226,8 @@ mov r29 = ar.bsp - st8 [r31] = r27, -8 ;; /* save ar.rnat */ - st8 [r31] = r30, -8 ;; /* save new value written to ar.bspstore */ - st8 [r31] = r28, -8 ;; /* save ar.bspstore */ - st8 [r31] = r29, -8 /* save ar.bsp */ + st8 [r31] = r27, -STACK_ITEM_SIZE ;; /* save ar.rnat */ + st8 [r31] = r30, -STACK_ITEM_SIZE ;; /* save new value written to ar.bspstore */ + st8 [r31] = r28, -STACK_ITEM_SIZE ;; /* save ar.bspstore */ + st8 [r31] = r29, -STACK_ITEM_SIZE /* save ar.bsp */ mov ar.rsc = r24 /* restore RSE's setting + kernel privileges */ @@ -246,9 +241,9 @@ cover /* allocate zero size frame (step 1 (from Intel Docs)) */ - add r31 = (STACK_SCRATCH_AREA_SIZE + (FRS_TO_SAVE * 2 * 8)), r12 ;; - - ld8 r30 = [r31], +8 ;; /* load ar.bsp */ - ld8 r29 = [r31], +8 ;; /* load ar.bspstore */ - ld8 r28 = [r31], +8 ;; /* load ar.bspstore_new */ + add r31 = STACK_SCRATCH_AREA_SIZE + ISTATE_OFFSET_AR_BSP, r12 ;; + + ld8 r30 = [r31], +STACK_ITEM_SIZE ;; /* load ar.bsp */ + ld8 r29 = [r31], +STACK_ITEM_SIZE ;; /* load ar.bspstore */ + ld8 r28 = [r31], +STACK_ITEM_SIZE ;; /* load ar.bspstore_new */ sub r27 = r30 , r28 ;; /* calculate loadrs (step 2) */ shl r27 = r27, 16 @@ -261,8 +256,8 @@ loadrs /* (step 3) */ - ld8 r27 = [r31], +8 ;; /* load ar.rnat */ - ld8 r26 = [r31], +8 ;; /* load cr.ifs */ - ld8 r25 = [r31], +8 ;; /* load ar.pfs */ - ld8 r24 = [r31], +8 ;; /* load ar.rsc */ + ld8 r27 = [r31], +STACK_ITEM_SIZE ;; /* load ar.rnat */ + ld8 r26 = [r31], +STACK_ITEM_SIZE ;; /* load cr.ifs */ + ld8 r25 = [r31], +STACK_ITEM_SIZE ;; /* load ar.pfs */ + ld8 r24 = [r31], +STACK_ITEM_SIZE ;; /* load ar.rsc */ mov ar.bspstore = r29 ;; /* (step 4) */ @@ -275,9 +270,9 @@ /* 17. restore interruption state from memory stack */ - ld8 r28 = [r31], +8 ;; /* load cr.ifa */ - ld8 r27 = [r31], +8 ;; /* load cr.isr */ - ld8 r26 = [r31], +8 ;; /* load cr.iipa */ - ld8 r25 = [r31], +8 ;; /* load cr.ipsr */ - ld8 r24 = [r31], +8 ;; /* load cr.iip */ + ld8 r28 = [r31], +STACK_ITEM_SIZE ;; /* load cr.ifa */ + ld8 r27 = [r31], +STACK_ITEM_SIZE ;; /* load cr.isr */ + ld8 r26 = [r31], +STACK_ITEM_SIZE ;; /* load cr.iipa */ + ld8 r25 = [r31], +STACK_ITEM_SIZE ;; /* load cr.ipsr */ + ld8 r24 = [r31], +STACK_ITEM_SIZE ;; /* load cr.iip */ mov cr.iip = r24;; @@ -292,13 +287,13 @@ and r25 = r25, r27 and r24 = r24, r26 ;; - or r25 = r25, r24;; + or r25 = r25, r24 ;; mov cr.ipsr = r25 /* 18. restore predicate registers from memory stack */ - ld8 r29 = [r31], +8 ;; /* load predicate registers */ + ld8 r29 = [r31], +STACK_ITEM_SIZE ;; /* load predicate registers */ mov pr = r29 /* 19. return from interruption */ - ld8 r12 = [r31] /* load stack pointer */ + ld8 r12 = [r31] /* load stack pointer */ rfi ;; @@ -373,46 +368,46 @@ mov loc46 = r31 - add r24 = 96 + STACK_SCRATCH_AREA_SIZE, r12 - add r25 = 112 + STACK_SCRATCH_AREA_SIZE, r12 - add r26 = 0 + STACK_SCRATCH_AREA_SIZE, r12 - add r27 = 16 + STACK_SCRATCH_AREA_SIZE, r12 - add r28 = 32 + STACK_SCRATCH_AREA_SIZE, r12 - add r29 = 48 + STACK_SCRATCH_AREA_SIZE, r12 - add r30 = 64 + STACK_SCRATCH_AREA_SIZE, r12 - add r31 = 80 + STACK_SCRATCH_AREA_SIZE, r12 ;; - - stf.spill [r26] = f2, 0x80 - stf.spill [r27] = f3, 0x80 - stf.spill [r28] = f4, 0x80 - stf.spill [r29] = f5, 0x80 - stf.spill [r30] = f6, 0x80 - stf.spill [r31] = f7, 0x80 ;; - - stf.spill [r24] = f8, 0x80 - stf.spill [r25] = f9, 0x80 - stf.spill [r26] = f10, 0x80 - stf.spill [r27] = f11, 0x80 - stf.spill [r28] = f12, 0x80 - stf.spill [r29] = f13, 0x80 - stf.spill [r30] = f14, 0x80 - stf.spill [r31] = f15, 0x80 ;; - - stf.spill [r24] = f16, 0x80 - stf.spill [r25] = f17, 0x80 - stf.spill [r26] = f18, 0x80 - stf.spill [r27] = f19, 0x80 - stf.spill [r28] = f20, 0x80 - stf.spill [r29] = f21, 0x80 - stf.spill [r30] = f22, 0x80 - stf.spill [r31] = f23, 0x80 ;; - - stf.spill [r24] = f24, 0x80 - stf.spill [r25] = f25, 0x80 - stf.spill [r26] = f26, 0x80 - stf.spill [r27] = f27, 0x80 - stf.spill [r28] = f28, 0x80 - stf.spill [r29] = f29, 0x80 - stf.spill [r30] = f30, 0x80 - stf.spill [r31] = f31, 0x80 ;; + add r24 = ISTATE_OFFSET_F8 + STACK_SCRATCH_AREA_SIZE, r12 + add r25 = ISTATE_OFFSET_F9 + STACK_SCRATCH_AREA_SIZE, r12 + add r26 = ISTATE_OFFSET_F2 + STACK_SCRATCH_AREA_SIZE, r12 + add r27 = ISTATE_OFFSET_F3 + STACK_SCRATCH_AREA_SIZE, r12 + add r28 = ISTATE_OFFSET_F4 + STACK_SCRATCH_AREA_SIZE, r12 + add r29 = ISTATE_OFFSET_F5 + STACK_SCRATCH_AREA_SIZE, r12 + add r30 = ISTATE_OFFSET_F6 + STACK_SCRATCH_AREA_SIZE, r12 + add r31 = ISTATE_OFFSET_F7 + STACK_SCRATCH_AREA_SIZE, r12 ;; + + stf.spill [r26] = f2, 8 * FLOAT_ITEM_SIZE + stf.spill [r27] = f3, 8 * FLOAT_ITEM_SIZE + stf.spill [r28] = f4, 8 * FLOAT_ITEM_SIZE + stf.spill [r29] = f5, 8 * FLOAT_ITEM_SIZE + stf.spill [r30] = f6, 8 * FLOAT_ITEM_SIZE + stf.spill [r31] = f7, 8 * FLOAT_ITEM_SIZE ;; + + stf.spill [r24] = f8, 8 * FLOAT_ITEM_SIZE + stf.spill [r25] = f9, 8 * FLOAT_ITEM_SIZE + stf.spill [r26] = f10, 8 * FLOAT_ITEM_SIZE + stf.spill [r27] = f11, 8 * FLOAT_ITEM_SIZE + stf.spill [r28] = f12, 8 * FLOAT_ITEM_SIZE + stf.spill [r29] = f13, 8 * FLOAT_ITEM_SIZE + stf.spill [r30] = f14, 8 * FLOAT_ITEM_SIZE + stf.spill [r31] = f15, 8 * FLOAT_ITEM_SIZE ;; + + stf.spill [r24] = f16, 8 * FLOAT_ITEM_SIZE + stf.spill [r25] = f17, 8 * FLOAT_ITEM_SIZE + stf.spill [r26] = f18, 8 * FLOAT_ITEM_SIZE + stf.spill [r27] = f19, 8 * FLOAT_ITEM_SIZE + stf.spill [r28] = f20, 8 * FLOAT_ITEM_SIZE + stf.spill [r29] = f21, 8 * FLOAT_ITEM_SIZE + stf.spill [r30] = f22, 8 * FLOAT_ITEM_SIZE + stf.spill [r31] = f23, 8 * FLOAT_ITEM_SIZE ;; + + stf.spill [r24] = f24 + stf.spill [r25] = f25 + stf.spill [r26] = f26 + stf.spill [r27] = f27 + stf.spill [r28] = f28 + stf.spill [r29] = f29 + stf.spill [r30] = f30 + stf.spill [r31] = f31 ;; mov loc47 = ar.fpsr /* preserve floating point status register */ @@ -442,46 +437,46 @@ /* 13. restore general and floating-point registers */ - add r24 = 96 + STACK_SCRATCH_AREA_SIZE, r12 - add r25 = 112 + STACK_SCRATCH_AREA_SIZE, r12 - add r26 = 0 + STACK_SCRATCH_AREA_SIZE, r12 - add r27 = 16 + STACK_SCRATCH_AREA_SIZE, r12 - add r28 = 32 + STACK_SCRATCH_AREA_SIZE, r12 - add r29 = 48 + STACK_SCRATCH_AREA_SIZE, r12 - add r30 = 64 + STACK_SCRATCH_AREA_SIZE, r12 - add r31 = 80 + STACK_SCRATCH_AREA_SIZE, r12 ;; - - ldf.fill f2 = [r26], 0x80 - ldf.fill f3 = [r27], 0x80 - ldf.fill f4 = [r28], 0x80 - ldf.fill f5 = [r29], 0x80 - ldf.fill f6 = [r30], 0x80 - ldf.fill f7 = [r31], 0x80 ;; - - ldf.fill f8 = [r24], 0x80 - ldf.fill f9 = [r25], 0x80 - ldf.fill f10 = [r26], 0x80 - ldf.fill f11 = [r27], 0x80 - ldf.fill f12 = [r28], 0x80 - ldf.fill f13 = [r29], 0x80 - ldf.fill f14 = [r30], 0x80 - ldf.fill f15 = [r31], 0x80 ;; - - ldf.fill f16 = [r24], 0x80 - ldf.fill f17 = [r25], 0x80 - ldf.fill f18 = [r26], 0x80 - ldf.fill f19 = [r27], 0x80 - ldf.fill f20 = [r28], 0x80 - ldf.fill f21 = [r29], 0x80 - ldf.fill f22 = [r30], 0x80 - ldf.fill f23 = [r31], 0x80 ;; - - ldf.fill f24 = [r24], 0x80 - ldf.fill f25 = [r25], 0x80 - ldf.fill f26 = [r26], 0x80 - ldf.fill f27 = [r27], 0x80 - ldf.fill f28 = [r28], 0x80 - ldf.fill f29 = [r29], 0x80 - ldf.fill f30 = [r30], 0x80 - ldf.fill f31 = [r31], 0x80 ;; + add r24 = ISTATE_OFFSET_F8 + STACK_SCRATCH_AREA_SIZE, r12 + add r25 = ISTATE_OFFSET_F9 + STACK_SCRATCH_AREA_SIZE, r12 + add r26 = ISTATE_OFFSET_F2 + STACK_SCRATCH_AREA_SIZE, r12 + add r27 = ISTATE_OFFSET_F3 + STACK_SCRATCH_AREA_SIZE, r12 + add r28 = ISTATE_OFFSET_F4 + STACK_SCRATCH_AREA_SIZE, r12 + add r29 = ISTATE_OFFSET_F5 + STACK_SCRATCH_AREA_SIZE, r12 + add r30 = ISTATE_OFFSET_F6 + STACK_SCRATCH_AREA_SIZE, r12 + add r31 = ISTATE_OFFSET_F7 + STACK_SCRATCH_AREA_SIZE, r12 ;; + + ldf.fill f2 = [r26], 8 * FLOAT_ITEM_SIZE + ldf.fill f3 = [r27], 8 * FLOAT_ITEM_SIZE + ldf.fill f4 = [r28], 8 * FLOAT_ITEM_SIZE + ldf.fill f5 = [r29], 8 * FLOAT_ITEM_SIZE + ldf.fill f6 = [r30], 8 * FLOAT_ITEM_SIZE + ldf.fill f7 = [r31], 8 * FLOAT_ITEM_SIZE ;; + + ldf.fill f8 = [r24], 8 * FLOAT_ITEM_SIZE + ldf.fill f9 = [r25], 8 * FLOAT_ITEM_SIZE + ldf.fill f10 = [r26],8 * FLOAT_ITEM_SIZE + ldf.fill f11 = [r27], 8 * FLOAT_ITEM_SIZE + ldf.fill f12 = [r28], 8 * FLOAT_ITEM_SIZE + ldf.fill f13 = [r29], 8 * FLOAT_ITEM_SIZE + ldf.fill f14 = [r30], 8 * FLOAT_ITEM_SIZE + ldf.fill f15 = [r31], 8 * FLOAT_ITEM_SIZE ;; + + ldf.fill f16 = [r24], 8 * FLOAT_ITEM_SIZE + ldf.fill f17 = [r25], 8 * FLOAT_ITEM_SIZE + ldf.fill f18 = [r26], 8 * FLOAT_ITEM_SIZE + ldf.fill f19 = [r27], 8 * FLOAT_ITEM_SIZE + ldf.fill f20 = [r28], 8 * FLOAT_ITEM_SIZE + ldf.fill f21 = [r29], 8 * FLOAT_ITEM_SIZE + ldf.fill f22 = [r30], 8 * FLOAT_ITEM_SIZE + ldf.fill f23 = [r31], 8 * FLOAT_ITEM_SIZE ;; + + ldf.fill f24 = [r24] + ldf.fill f25 = [r25] + ldf.fill f26 = [r26] + ldf.fill f27 = [r27] + ldf.fill f28 = [r28] + ldf.fill f29 = [r29] + ldf.fill f30 = [r30] + ldf.fill f31 = [r31] ;; mov r1 = loc17